[BioC] Advice with RemoveBatchEffect and Rank Product package

[Gordon K Smyth]

Dear Osee,

You are attempting to do a number of things that don't seem correct to me.

First, you seem to attempting a separate channel analysis of two color 
microarray data, but ignoring the pairing of the red and green channels. 
It isn't correct to do this.  I don't see any way to use RankProd, or any 
other package designed for independent samples, correctly in this context. 
If you must do a separate channel analysis, you would be better off using 
the separate channel analysis facilities of the limma package.

Second, when you set batch=rep(1,24), you are defining a batch that 
consists of every array in your data set.  Obviously it doesn't make sense 
to remove batch effects unless there are at least two batches.

Third, I don't follow your design matrix.

It would be better to go back to the start, and describe in more basic 
terms what is the nature of your data and what comparison you are trying 
to make.

Best wishes
Gordon

> Date: Sat, 8 Sep 2012 11:40:45 +0000
> From: "Sanogo, Yibayiri O" <sanogo at illinois.edu>
> To: "bioconductor at r-project.org" <bioconductor at r-project.org>
> Subject: [BioC] Advice with RemoveBatchEffect and Rank Product package
>
> Dear Members of the list,
>
> (I apologize for posting this again -I sent it earlier to the list but 
> from another account and I was listed me as non-Member -and Member I am 
> since 2008:-)).
>
> I have been using Rank Prod to analyze Agilent two-color data. However, I
> would like to remove the dye effect prior to analysis. I read on the forum
> that RemoveBatchEffect should be used in the Limma linear model, a type of
> modeling that is not in Rank Product.
>
> I have two questions:
>
> 1) Would it be appropriate to use RemoveBatchEffect to correct for dye
> effect prior to running the expression data using Rank Prod?
>
> 2) a) If no, what other function could I use to do this?
>   b) If yes, I would like a help with the correct design and how to
> properly indicate the batch.
>
> Here is my design indicating the two dyes (cy3=-1, cy5=1; T, D, C, BS =are
> different areas of the brain):
>
> design1
>   BS  C  D  T
> 1   0  0  0  1
> 2   0  0  0 -1
> 3   0  0  0  1
> 4   0  0  0 -1
> 5   0  0  0  1
> 6   0  0  0 -1
> 7   0  0 -1  0
> 8   0  0  1  0
> 9   0  0 -1  0
> 10  0  0  1  0
> 11  0  0 -1  0
> 12  0  0  1  0
> 13  0  1  0  0
> 14  0 -1  0  0
> 15  0  1  0  0
> 16  0 -1  0  0
> 17  0  1  0  0
> 18  0 -1  0  0
> 19 -1  0  0  0
> 20  1  0  0  0
> 21 -1  0  0  0
> 22  1  0  0  0
> 23 -1  0  0  0
> 24  1  0  0  0
>
> attr(,"assign")
> [1] 1 1 1 1
>
> I've tried this (Data_RP are my data, the M values of the expression set):
>
> DYE_RP<-removeBatchEffect(Data_RP, batch=rep(1,24), batch2=NULL,
> design=design1)
>
> but it is returning an error message
> " Error in contr.sum(levels(batch)) :
>  not enough degrees of freedom to define contrasts"
>
> Please help me correct this code.
>
> Thank you so much for your help.
>
> Osee
>
> -- -- --
> Y. Osee Sanogo
> Integrative Biology
> Institute for Genomic Biology
> University of Illinois at Urbana
> 505 S. Goodwin Ave
> Urbana, IL-61801
>
> Tel: 217-333 2308 (Office)
>     217-417 9593 (Cell)

______________________________________________________________________
The information in this email is confidential and intend...{{dropped:4}}